Accelerating Fragment Based Drug Discovery using Grand Canonical Nonequilibrium Candidate Monte Carlo

Fragment-based drug discovery is a popular approach in academia and industry for the early stages of drug development. Computational tools have become integral to these campaigns and provide a route to library design, virtual screening, the identification of putative small molecule binding sites, the elucidation of binding geometries, and the prediction of accurate binding affinities. Molecular dynamics-based simulations have become increasingly popular, but are often limited by sampling issues related to the simulation timescales obtainable. Here, we expand the use of grand canonical nonequilibrium candidate Monte Carlo (GCNCMC) to overcome these limitations and accurately predict the binding sites, modes, and affinities of fragment-like molecules. GCNCMC has been used previously to accurately predict the location of water molecules in protein-ligand systems, by attempting the insertion and deletion of water to, or from, a region of interest; each proposed move is subject to a rigorous acceptance test based on the thermodynamic properties of the system. Here, we demonstrate the ability of fragment-based GCNCMC to rapidly and reliably find occluded experimental fragment binding sites. We also show that the method can accurately sample multiple fragment binding modes without any prior knowledge of their existence. Finally, we calculate the binding affinities for fragment molecules to three systems. We find that our results are in agreement with a more established method, namely absolute binding free energy calculations. Notably, GCNCMC does not require the use of complex restraints, the handling of multiple binding modes, or post-hoc symmetry corrections. Rather, binding sites, geometries and affinities all arise naturally from a series of GCNCMC simulations.